Variable Trust Setting for Safe and Ethical Algorithms for Navigation of Autonomous Vehicles (C-NAV) on a Highway

This paper presents the use of an ethical model-to-decision approach for promoting safe manoeuvrability of autonomous vehicles (AVs) on highways, when considering scenarios such as exiting a highway via a slip road. In this research, a modelling and simulation approach is undertaken. The modelling involves the use of an adaptive model-predictive control (MPC) algorithm with a dynamic bicycle model. The approach was developed to incorporate a novel continuous evaluation of the distances between AVs (considering virtual boundaries), logical sequences towards achieving safe lane change and slip road exit manoeuvres (driving rules based on deontological ethics), and control logic towards accounting for acceleration, deceleration, and constant velocity. Based on this, a novel continuous risk assessment algorithm has been developed based on the product of collision probability and harm. This has been used to investigate the introduction of a novel trust setting that gives the user ‘control ’ of how the AV operates around other AVs. The results presented in the paper highlight the effectiveness of the approach, i.e., the ability to undertake ethical and safe manoeuvres in the event of difficult highway decision scenarios such as slip road exits.